Object identification system with reduced service calls

Abstract

Methods and systems for implementing an object identification system equipped to reduce the number of service calls for remote object identification lookups are disclosed. The object identification management server works to reduce the number of object identification requests, which overall improves the process of quickly and accurately providing object identifications when needed and without unnecessarily wasting computing resources. The object identification server can subsequently manage the number of identification requests by setting a tunable threshold, which allows a system to operate more efficiently and allocate computing resources accordingly.

We claim: 
     
       1. An object identification management server comprising:
 a processor; 
 a memory operatively storing instructions which, when executed by the processor, implement an object identification service configured to:
 execute an object detection algorithm on each frame of video content to detect objects that appear within the frame; 
 for each frame in which an object is detected:
 determine whether the object corresponds to an identified object present in a preceding frame using an object tracking algorithm; 
 based at least in part on a determination that the object is not identified in a preceding frame:
 call a remote object identification interface to obtain an identification of the object; and 
 based on receiving the identification of the object, 
 associate the identification of the object with the object in the frame; and 
 
 based at least in part on a determination that the object was present and identified in the preceding frame, associate an identification of the object from the preceding frame with the object in the frame without requiring the object identification service to call the remote object identification interface to identify the object in the frame. 
 
 
 
     
     
       2. The object identification management server of  claim 1 , wherein the object is a face of a person, and the remote object identification interface provides a facial identification service. 
     
     
       3. The object identification management server of  claim 2 , wherein calling the remote object identification interface includes transmitting a request to a remote API. 
     
     
       4. The object identification management server of  claim 3 , further comprising receiving a setting of a tunable threshold, wherein the tunable threshold adjusts the number of identification requests transmitted to the remote object identification interface. 
     
     
       5. The object identification management server of  claim 4 , wherein the tunable threshold determines a maximum time length between requests sent to the remote object identification interface. 
     
     
       6. The object identification management server of  claim 2 , wherein the object identification service is further configured to associate metadata with each frame of the video content, the metadata including an identification of the person and a confidence level in the identification of the person. 
     
     
       7. The object identification management server of  claim 2 , wherein the object identification service receives a unique identification for each person identified in the video content from the remote object identification interface. 
     
     
       8. The object identification management server of  claim 2 , wherein the object identification service is further configured to provide to the remote object identification interface one or more identities, the one or more identities, including a representative image of a person and a label of the person. 
     
     
       9. The object identification management server of  claim 8 , wherein the label of the person comprises a name of the person. 
     
     
       10. The object identification management server of  claim 1 , wherein the object identification service is further configured to receive the video content from one or more real-time video streams. 
     
     
       11. The object identification management server of  claim 1 , wherein the one or more real-time video streams are received from security cameras. 
     
     
       12. A facial identification system useable to identify persons appearing within video content, the object identification system comprising:
 a computing system having a communication interface, a memory, and a processor, the memory storing instructions which, when executed by the processor, implement a facial identification service configured to:
 execute an object detection algorithm on each frame of video content to detect objects that appear within the frame, the objects including at least one face object representative of a face of a person; 
 for each frame:
 based on detection of a face object within the frame, determine whether the face object corresponds to an identified face object present in a preceding frame using an object tracking algorithm; 
 based at least in part on a determination that the face object is not identified in the preceding frame:
 call a remote facial identification interface to obtain an identification of a person based on the face object; and 
 based on receiving the identification of the person, associate the identification of the person with the face object; and 
 
 based at least in part on a determination that the face object was present and identified in the preceding frame, associate the face object with the identification of the person without requiring the remote facial identification interface to be called in response to detection of the face object within the frame. 
 
 
 
     
     
       13. The facial identification system of  claim 12 , wherein the object tracking algorithm comprises applying a Kalman filter to track object movements between frames of the video content. 
     
     
       14. The facial identification system of  claim 12 , wherein the object detection algorithm comprises a facial detection algorithm. 
     
     
       15. The facial identification system of using  claim 12 , wherein the facial identification service tracks a similarity between the frame and the preceding frame in an area of the frame, including the at least one face object. 
     
     
       16. The facial identification system of using  claim 12 , wherein the facial identification service is configured to receive, from the remote facial identification interface, a confidence level associated with a closest facial identification result. 
     
     
       17. The facial identification system of using  claim 12 , wherein the facial identification service is configured to transmit a new identification request at user-definable time increments. 
     
     
       18. The facial identification system of using  claim 12 , wherein the facial identification service periodically transmits a new identification request to the remote facial identification interface irrespective of whether the face object is identified in the preceding frame. 
     
     
       19. A computer-implemented method for transferring data from an entity to a processing server comprising:
 receiving video data generated from the entity's video capture system; 
 comparing accessible records owned by the entity provided to the third-party from previously provided photos; 
 performing an object detection algorithm on each frame of video content for detecting objects appearing within the frame, the objects including at least one face object representative of a face of a person;
 for each frame:
 based on detection of a face object within the frame, determining whether the face object corresponds to an identified face object present in a preceding frame using an object tracking algorithm; 
 based at least in part on a determination that the face object is not identified in the preceding frame:
 calling a remote facial identification interface to obtain an identification of a person based on the face object; and 
 based on receiving the identification of the person, associate the identification of the person with the face object; and 
 
 based at least in part on a determination that the face object was present and identified in the preceding frame, associate the face object with the identification of the person without requiring the remote facial identification interface to be called in response to the detection of the face object within the frame. 
 
 
 
     
     
       20. The computer-implemented method for transferring data from an entity to the processing server of  claim 19 , wherein a client's mobile device is provided modified video data including the initial video data with an overlay identifying at least one person appearing in the video data based on the identification received from the remote facial identification interface. 
     
     
       21. The computer-implemented method for transferring data from an entity to the processing server of  claim 19 , wherein a client's mobile device is provided modified video data including the initial video data with an overlay identifying at least one person appearing in the video data that is unidentified based on the lack of identification received from the remote facial identification interface. 
     
     
       22. The computer-implemented method for transferring data from an entity to the processing server of  claim 19 , wherein video data is at least displayed on the client's mobile device remotely from the processing server.